{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:27PXWOTD5QFNBYNROIKNZDABUK","short_pith_number":"pith:27PXWOTD","schema_version":"1.0","canonical_sha256":"d7df7b3a63ec0ad0e1b17214dc8c01a29d656065f0a564340375f314573fed5b","source":{"kind":"arxiv","id":"2402.02030","version":2},"attestation_state":"computed","paper":{"title":"Panacea: Pareto Alignment via Preference Adaptation for LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chengdong Ma, Haojun Chen, Qingfu Zhang, Siyuan Qi, Xiaoyuan Zhang, Yaodong Yang, Yifan Zhong, Ziran Yang","submitted_at":"2024-02-03T05:01:04Z","abstract_excerpt":"Current methods for large language model alignment typically use scalar human preference labels. However, this convention tends to oversimplify the multi-dimensional and heterogeneous nature of human preferences, leading to reduced expressivity and even misalignment. This paper presents Panacea, an innovative approach that reframes alignment as a multi-dimensional preference optimization problem. Panacea trains a single model capable of adapting online and Pareto-optimally to diverse sets of preferences without the need for further tuning. A major challenge here is using a low-dimensional pref"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2402.02030","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-03T05:01:04Z","cross_cats_sorted":[],"title_canon_sha256":"fba7d1055cc6a2b66a17088d63cb208bd242c2c37eb233cf110ba9dce21b10bd","abstract_canon_sha256":"3f5aaa873e8d3d7c680257b99925f65f4e17e2e86754f490912488c8cc11a225"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:22:22.310244Z","signature_b64":"1lxt/nB7smO81KmZrWzXpg3XjSEggcWqIXENIKCSHGbQLZJ0f2+PIsn0HBdMuu1yFc3CZC7TF/DeCeE1AwxlCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d7df7b3a63ec0ad0e1b17214dc8c01a29d656065f0a564340375f314573fed5b","last_reissued_at":"2026-07-05T08:22:22.309677Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:22:22.309677Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Panacea: Pareto Alignment via Preference Adaptation for LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chengdong Ma, Haojun Chen, Qingfu Zhang, Siyuan Qi, Xiaoyuan Zhang, Yaodong Yang, Yifan Zhong, Ziran Yang","submitted_at":"2024-02-03T05:01:04Z","abstract_excerpt":"Current methods for large language model alignment typically use scalar human preference labels. However, this convention tends to oversimplify the multi-dimensional and heterogeneous nature of human preferences, leading to reduced expressivity and even misalignment. This paper presents Panacea, an innovative approach that reframes alignment as a multi-dimensional preference optimization problem. Panacea trains a single model capable of adapting online and Pareto-optimally to diverse sets of preferences without the need for further tuning. A major challenge here is using a low-dimensional pref"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02030","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2402.02030/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2402.02030","created_at":"2026-07-05T08:22:22.309736+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.02030v2","created_at":"2026-07-05T08:22:22.309736+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02030","created_at":"2026-07-05T08:22:22.309736+00:00"},{"alias_kind":"pith_short_12","alias_value":"27PXWOTD5QFN","created_at":"2026-07-05T08:22:22.309736+00:00"},{"alias_kind":"pith_short_16","alias_value":"27PXWOTD5QFNBYNR","created_at":"2026-07-05T08:22:22.309736+00:00"},{"alias_kind":"pith_short_8","alias_value":"27PXWOTD","created_at":"2026-07-05T08:22:22.309736+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.07162","citing_title":"CLIPer: Tailoring Diverse User Preference via Classifier-Guided Inference-Time Personalization","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK","json":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK.json","graph_json":"https://pith.science/api/pith-number/27PXWOTD5QFNBYNROIKNZDABUK/graph.json","events_json":"https://pith.science/api/pith-number/27PXWOTD5QFNBYNROIKNZDABUK/events.json","paper":"https://pith.science/paper/27PXWOTD"},"agent_actions":{"view_html":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK","download_json":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK.json","view_paper":"https://pith.science/paper/27PXWOTD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.02030&json=true","fetch_graph":"https://pith.science/api/pith-number/27PXWOTD5QFNBYNROIKNZDABUK/graph.json","fetch_events":"https://pith.science/api/pith-number/27PXWOTD5QFNBYNROIKNZDABUK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK/action/storage_attestation","attest_author":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK/action/author_attestation","sign_citation":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK/action/citation_signature","submit_replication":"https://pith.science/pith/27PXWOTD5QFNBYNROIKNZDABUK/action/replication_record"}},"created_at":"2026-07-05T08:22:22.309736+00:00","updated_at":"2026-07-05T08:22:22.309736+00:00"}